Comparative Analysis of PID-Driven Data-Based and PSO-Tuned Fuzzy Membership Functions for Robotic Manipulator Control

Abstract

Robotic manipulators require control systems that are both responsive and precise in order to ensure accurate tracking and stability in dynamic environments. Conventional fuzzy logic controllers that are based on proportional integral derivative (PID) methods frequently encounter difficulties in achieving fast response, minimal steady-state error, and low overshoot. This study presents a comparative evaluation of a PID-driven data-based fuzzy logic controller and a particle swarm optimization (PSO) tuned fuzzy logic controller for a three-axis robotic manipulator implemented in Simulink. Both controllers used Gaussian membership functions within a Mamdani inference structure. The PSO algorithm was employed to optimize fuzzy input-scaling gains using a composite performance index that incorporated absolute error, control effort, overshoot penalty, and steady-state error. The simulation results indicate that the PSO-tuned controller consistently outperformed the benchmark. On the R-axis, it shortened rise and settling times and reduced overshoot, mean absolute error (MAE), and root mean square error (RMSE). On the T-axis, response speed and error values improved, although overshoot increased, indicating a trade-off between speed and stability. On the Z-axis, the PSO controller achieved a substantial decrease in overshoot, lower error metrics, and faster stabilization. Overall, the PSO-based tuning process preserved steady-state stability while improving transient performance on all axes. These findings show that metaheuristic optimization is an effective and practical method for enhancing fuzzy logic controllers in robotic manipulators. This approach has potential applications in precision manufacturing, service automation, and surgical robotics.